Many tasks in computer vision involve assigning a label, such as disparity for depth of field data to every pixel. Energy minimization may be used to accomplish this labeling. The present invention provides an efficient way of minimizing energy functions in assigning labels to pixels. Two methods of using graph cuts to compute a local minimum are described. The first move is an αβ swap. For a pair of labels, α, β, this move swaps the labels between an arbitrary set of pixels labeled a and another arbitrary set of pixels labeled β. The first method generates a labeling such that there is no swap move that decreases the energy. The second move is the α-expansion. For a label α, this move assigns an arbitrary set of pixels with the label α.
|
1. A process for determining an optimal labeling of pixels in early vision, comprising the steps of:
a) providing an initial solution having a plurality of labels, wherein each pixel from a plurality of pixels has a corresponding pixel label from the plurality of labels; b) forming a graph from said initial solution, said graph comprised of nodes and of edges connecting said nodes, each said pixel having a corresponding node; c) selecting a move from said graph, wherein said move comprises assigning a new corresponding pixel label to at least one pixel from said plurality of pixels; d) assigning a weight value to each said edge of said; e) determining a minimum graph cur in response to said assigned weights, wherein a graph cut comprises a subset of said edges; a minimum graph cut having a smallest value of a cost; f) forming a graph from results of said minimum graph cut; and g) repeating steps c-f until predetermined criterion is reached.
2. The process of
3. The process of
4. The process of
5. The process of
6. The process of
|
This application claims priority of U.S. provisional applications Serial No. 60/151,413 entitled, "Fast Approximate Energy Minimization Via Graph Cuts" filed Aug. 30, 1999 by the present applicants.
This invention was partially funded by the Government under a grant from DARPA under contract DAAL01-97-K-0104, and by NSF awards CDA-9703470 and IIS-9900115. The Government has certain rights in portions of the invention.
This invention relates generally to computer vision and more particularly to minimizing energy functions in labeling pixels in early vision.
In computer vision, early vision is usually considered to involve the description of geometric structure in an image or sequence of images. The behavioral constraints on real-time visual systems typically require that the early vision stage be fast, reliable, general and automatic. Many early vision problems require estimating some spatially varying quantity such as intensity or disparity, from noisy measurements.
Spatially varying quantities tend to be piecewise smooth, i.e. they vary smoothly at most points, but change dramatically at object boundaries. Every pixel in a set P must be assigned a label in some set L. For motion or stereo, the labels are disparities. For image restoration, the labels represent intensities. The goal is to find a labeling f that assigns each pixel p∈P a label fp∈L, where f is both piecewise smooth and consistent with observed data.
Computer vision problems can be formulated in terms of minimization of energy functions. Energy functions, however, are generally difficult to minimize. The major difficulty with energy minimization for computer vision lies in the enormous computational costs. Energy functions for computer vision typically have many local minima. Also, the space of possible labelings has the dimension |P| which is many thousands.
There have been many attempts to design fast algorithms for energy minimization. Due to the inefficiency of computing a global minimum, some solutions are directed instead at computing a local minimum. In general, a local minimum can be arbitrarily far from the optimum. It thus may not convey any of the global image properties that are encoded in the energy function. It is, however, difficult to determine the exact cause of an algorithmi's failures. When an algorithm gives unsatisfactory results, it may be due to either a poor choice of the energy function, or due to the fact that the answer is far form the global minimum. Local minimization techniques are also naturally sensitive to an initial estimate.
In general, a labeling f is a local minimum of the energy E if
In case of discrete labeling, the labelings near to f are those that lie within a single move of f. Many local optimization techniques use standard moves, where only one pixel can change its label at a time. For standard moves, the equation above can be read as follows: if you are at a local minimum with respect to standard moves, then you cannot decrease the energy by changing a single pixel's label. In fact, this is a very weak condition. As a result, optimization schemes using standard moves frequently generate low quality solutions.
An example of a local method using standard moves is Iterated Conditional Modes (ICM). For each site (pixel or voxel), the label which gives the largest decrease of the energy function is chosen, until the iteration converges to a local minimum.
Another example of an algorithm using standard moves is simulated annealing. Simulated annealing randomly proposes some change in the state of the system. If the change results in a decrease of energy (which is equivalent to a decrease in cost in the more general sense of optimization), the change will always be taken. If it results in an increase in energy, it will be chosen using a probability scheme. At high temperatures (i.e. early in the simulated annealing process), changes of state that result in large increases in energy will be accepted with a higher probability than they would be at low temperatures (late in the simulated annealing process). Simulated annealing is widely used because it can optimize an arbitrary energy function. Minimizing an arbitrary energy function requires exponential time, and as a consequence simulated annealing is very slow.
Simulated annealing is inefficient partly because at each step, annealing changes the value of a single pixel. Theoretically, simulated annealing should eventually find the global minimum if it is run long enough. As a practical matter, it is necessary to decrease the algorithm's temperature parameter faster than required by the theoretically optimal schedule. Once annealing's temperature parameter is sufficiently low, the algorithm will converge to a local minimum with respect to standard moves.
An alternative method is to seek a local minimum using variational techniques, for example. Variational methods can be applied if the energy minimization problem is phrased in continuous terms. Variational techniques use Euler equations, which are guaranteed to hold at a local minimum. In continuous cases, the labels near to f in the equation above are normally defined as ∥f-f'∥≦ε where ε is a positive, constant and ∥·∥ is a norm, e.g. L2 over some appropriate functional space. To apply these algorithms to actual imagery requires discretization.
Another alternative is to use discrete relaxation labeling methods. In relaxation labeling, combinatorial optimization is converted into continuous optimization with linear constraints. Then some form of gradient descent, which gives the solution satisfying the constraints, is used.
There are also methods that have optimality guarantees in certain cases. Continuation methods, such as graduated non-convexity are an example. These methods involve approximating an intractable (non-convex) energy function by a sequence of energy functions beginning with a tractable (convex) approximation. There are circumstances where these methods are known to compute the optimal solution. Continuation methods can be applied to a large number of energy functions, but except for special cases, nothing is known about the quality of their output.
Mean field annealing is another popular minimization approach. It is based on estimating the partition function from which the minimum of the energy can be deduced. Computing the partition function, however, is computationally intractable, and saddle point approximations are used.
There are a few interesting energy functions where the global minimum can be rapidly computed via dynamic programming. Dynamic programming, however, is restricted essentially to energy functions in one-dimensional settings. In general, the two-dimensional energy functions that arise in early vision cannot be solved efficiently via dynamic programming.
It is an object of the present invention to provide a method and apparatus to improve the early vision stage of computer vision.
It is another object of the present invention to provide a method and apparatus to improve the speed while maintaining accuracy of minimizing energy functions.
The problems of early vision and minimizing energy functions are solved by the present invention of a method and apparatus for fast approximate energy minimization via graph cuts.
Many tasks in computer vision involve assigning a label, such as disparity for depth of field data to every pixel. Energy minimization may be used to accomplish this labeling. The present invention provides an efficient way of minimizing energy functions in assigning labels to pixels.
The major restriction is that the energy function's smoothness term must involve only pairs of pixels. Two methods of using graph cuts to compute a local minimum are described. They may be used even when very large moves are allowed.
The first move is an α-β swap. For a pair of labels, α, β, this move swaps the labels between an arbitrary set of pixels labeled a and another arbitrary set of pixels labeled β.
The first method generates a labeling such that there is no swap move that decreases the energy.
The second move is the α-expansion. For a label α, this move assigns an arbitrary set of pixels with the label α.
The second method which requires the smoothnes term to be a metric, generates a labeling such that there is no expansion move that decreases the energy.
The present invention together with the above and other advantages may best be understood from the following detailed description of the embodiments of the invention illustrated in the drawings, wherein:
In the algorithms of the present invention, an arbitrary initial solution is assumed. The initial solution may be completely arbitrary or it could be some less accurate solution to the computer vision problem then the present invention.
Energy functions are difficult and slow to solve. Approximate solutions provide good results in a reasonable period of time. The algorithms of the present invention provide approximate solutions with provable qualities. The algorithms of the present invention use graph cuts to obtain solutions. Each pixel in the view is a node. The graph can be thought of as a net and on the net are nodes and strings that connect the nodes. In a preferred embodiment of the invention, nodes are connected to neighboring nodes immediately above, below, to the right and to the left. The lines connecting the nodes represent the relationship between the nodes. In graph terminology, these lines are also referred to as "edges." Each line belongs to a distinct pair of pixels. This node and line system is also referred to as the "neighborhood system." In alternative embodiments of the invention, the graph could have diagonal links between nodes in addition to the links described above or instead of those links. Nodes could also be connected in ways other than to direct neighbors. Many types of neighborhood systems are possible and the invention is not limited to the neighborhood systems described herein.
Vp,q represents how pixels act within a pair of pixels. It is desirable for the difference between interacting pixels to be as small as possible. Vp,q can also be thought of as a penalty for discontinuity between pixels. The present system puts restrictions on Vp,q. Vp,q must be non-negative but it can be different for different pixel pairs. The penalties and restrictions are also part of the neighborhood system.
The input to the algorithms then is the neighborhood system, i.e. nodes, edges, penalty functions and assumptions.
The goal is to find a labeling f that minimizes the energy
where Esmooth measures the extent to which f is not piecewise smooth, and
Edata measures the disagreement between f and the observed data. The form of Edata measures the disagreement between f and the observed data. The form of Edata is typically,
where Dp measures how appropriate a label is for the pixel p given the observed data. In image restoration, for example, Dp(fp) is typically (fp-ip), where ip is the observed intensity of the pixel p.
The choice of Esmooth is important. There are many potential choices. For example, in regularization-based vision, Esmooth makes f smooth everywhere. This gives poor results at object boundaries in general. Energy functions that do not have this problem are called discontinuity-preserving.
The major difficulty with energy minimization for early vision lies in the enormous computational costs. Typically these energy functions have many local minimums (i.e. they are non-convex). Also, the space of the possible labelings has dimension |P|, which is many thousands.
The energy functions used in the present invention are of the form,
where N is the set of interacting pairs of pixels. Typically N consists of adjacent pixels, but it can be arbitrary. Dp is nonnegative but otherwise arbitrary. Only pairs of pixels interact directly in Esmooth. Each pair of pixels (p,q) can have its own distinct interaction penalty Vp,q independent of any other pair of interacting pixels. To simplify notation, Vp,q will be written as V.
The methods of the present invention approximately minimize the energy E(f) for an arbitrary finite set of labels L under two fairly general classes of interaction penalty V: metric and semi-metric: V is called a metric on the space of labels L if it satisfies:
for any labels α,β,γ∈L. If V does not satisfy the triangle inequality of equation (7), it is called semi-metric.
Both semi-metrics and metrics include important cases of discontinuity preserving interaction penalties. A penalty V is discontinuity preserving if it is bounded, i.e. sup(x,y)∈R
The methods of the present invention generate a local minimum with respect to very large moves, i.e. more than one pixel at a time. The algorithms generate a labeling that is a local minimum of the energy for two types of large moves: α-expansion and α-β-swap. In contrast to the standard moves described above in the Background of the Invention, these moves allow large numbers of pixels to change their labels simultaneously. This makes the set of labelings within a single move of a locally optimal f exponentially large. For example, α-expansion moves are so strong that any labeling locally optimal with respect to these moves are within a known factor of the global minimum.
Any labeling f can be uniquely represented by a partition of image pixels P={Pl|l∈L} where Pl={p∈P|fp=l} is a subset of pixels assigned label l. Since there is an obvious one to one correspondence between labelings f and partitions P, these notions will be used interchangingly.
Given a pair of labels α,β, a move from a partition P (labeling f) to a new partition P' (labeling f') is called an α-β swap if Pl=P'l for any label l≠α,β. This means that the only difference between P and P' is that some pixels that were labeled α in P are labeled a in P'. A special case of an α-β swap is a move that gives the label α to some set of pixels that used to be labeled β.
Given a label α, a move from a partition P (labeling f) to a new partition P' (labeling f') is called an α-expansion if Pα⊂P'αand P'l⊂Pl for any label l≠α. In other words, an α-expansion move allows any set of image pixels to change their labels to α.
A move which assigns a given label α to a single pixel is both an α-β swap and an α-expansion. As a consequence, a standard move is a special case of both an α-β swap and an α-expansion.
Algorithms and Properties
Referring again to
The algorithms are efficient graph based methods to find the optimal α-β or α-expansion given a labeling f. This is the key step in the algorithms. Once these methods are available, it is easy to design some variations of the "fastest descent" technique that can efficiently find the corresponding local minimums.
The structure of the algorithms is similar. A single execution of steps 3.1-3.2 is called an iteration, and an execution of steps 2-4 is a cycle. In each cycle, the algorithm performs an iteration for every label (expansion algorithm) or for every pair of labels (swap algorithm), in a certain order that can be fixed or random. A cycle is successful if a strictly better labeling is found at any iteration. The algorithms stop after the first unsuccessful cycle since no further improvement is possible. Obviously, a cycle in the swap algorithm takes |L|2 iterations, and a cycle in the expansion algorithm takes |L| iterations. These algorithms are guaranteed to terminate in a finite number of cycles. Termination can be proved to be in O(|P|) cycles. In experiments, the algorithm stops after a few cycles, and most of the improvements occur during the first cycle.
Graph cuts are used to efficiently find {circumflex over (f)} for the key part of each algorithm in step 3.1. Step 3.1 uses a single minimum cut on a graph whose size is O(|P|). The graph is dynamically updated after each iteration.
Graph Cuts
Before describing the key step 3.1 of the swap and the expansion algorithms, graph cuts will be reviewed. Let G=(V,ε) be a weighted graph with two distinguished vertices called the terminals. A cut C⊂ε is a set of edges such that the terminals are separated in the induced graph G(C)=(V,ε-C). In addition, no proper subset of C separates the terminals in G(C). The cost of the cut C, denoted |C|, equals the sum of its edge weights.
The minimum cut problem is to find the cut with smallest cost. This problem can be solved efficiently by computing the maximum flow between the terminals, according to a theorem due to Ford and Fulkerson. There are a large number of fast algorithms available in the art for this problem. The worst case complexity is low-order polynomial; however, for graphs of the present invention, the running time is nearly linear in practice.
Finding the Optimal Swap Move
Given an input labeling f (partition P) and a pair of labels α, β, a labeling {circumflex over (f)} needs to be found that minimizes E over all labelings within one α-β swap of f. This is the, critical step in the swap move algorithm of
This section is organized as follows. The construction of Gαβ for a given f (or P) is described. Cuts C on Gαβ correspond in a natural way to labelings fC which are within one α-β swap move of f. Theorem 4.4 shows that the cost of a cut is |C|=E(fC) plus a constant. A corollary from this theorem states the main result that the desired labeling {circumflex over (f)} equals fC where C is a minimum cut on Gαβ.
(the n-links). The weights assigned to the edges according to Table 1.
TABLE 1 | |||
edge | weight | for | |
tpα |
|
p ∈ Pαβ | |
tpβ |
|
p ∈ Pαβ | |
e{p,q} | V(α, β) | {p, q} ∈ N | |
p, q ∈ Pαβ | |||
Any cut C on Gαβ must sever (include) exactly one t-link for any pixel p∈Pαβ. If neither t-link were in C, there would be a path between the terminals; while if both t-links were cut, then a proper subset of C would be a cut. Thus, any cut leaves each pixel in Pαβ with exactly one t-link. This defines a natural labeling fC corresponding to a cut C on Gαβ,
In other words, if the pixel p is in Pαβ hen p is assigned label α when the cut C separates p from the terminal α; similarly, p is assigned label β when C separates p from the terminal β. If p is not in Pαβ then its initial label fp is retained. This implies
Lemma 4.1 A labeling fC corresponding to a cut C on Gαβ is one β-β swap away from the initial labeling f.
It is easy to show that a cut C severs an n-link e{p,q} between neighboring pixels on Gαβ if and only if C leaves the pixels p and q connected to different terminals.
Properties (a) and (b) follow from the requirement that no proper subset of C should separate the terminals. Properties (c) and (d) also use the fact that a cut has to separate the terminals.
The next lemma is a consequence of property 4.2 and equation 8.
Lemma 4.3 For any cut C and for any n-link e{p,q}
PROOF: There are four cases with similar structure; the case where tβα,tqβ∈C will be illustrated. In this case, e{p,q}∈C and, therefore, |C∩ε{p,q}|=V(α,β). By (8), fpC=α and fqeα=β.
Note that this proof assumes that V is a semi-metric, i.e. that equations 5 and 6 hold.
Lemmas 4.1 and 4.3 plus property 4.2 yield
Theorem 4.4 There is a one to one correspondence between cuts C on Gαβ and labelings hat are one α-β swap from f. Moreover, the cost of a cut C on Gαβ is |C|=E(fC) plus a constant.
Proof: The first part follows from the fact that the severed t-links uniquely determine the labels assigned to pixels p and the n-links that must be cut. The cost of a cut C is
Note that for p∈Pαβ,
Lemma 4.3 gives the second term in (10). Thus, the total cost of a cut C is
This can be rewritten as |C|=E(fC)-K where
is the same constant for all cuts C.
Corollary 4.5 The lowest energy labeling within a single α-β swap move from f is {circumflex over (f)}=fC, where C is the minimum cut on Gαβ.
Next, a pair of labels to swap is chosen, block 152. Then, using the graph cuts described above, the best swap for the chosen pair of labels is determined, block 154. The minimum cut is chosen by assigning weights to the various links in the graph according to Table 1. A minimum cut is calculated from the weighted links.
The new labeling that results from finding the best swap move is stored, block 156.
In the preferred embodiment of the invention, the method determines if any pair of label combinations can be improved, block 158. If yes, the algorithm moves on to another pair of labels and the process repeats until no label combinations can be improved. If there are no more improvements that can be made, the process ends, block 160. In alternative embodiments of the invention, the iteration continues only until some acceptable solution is obtained.
Finding the Optimal Expansion Move
Given an input labeling f (partition P) and a label α, a labeling {circumflex over (f)} that minimizes E over all labelings within one α-expansion of f needs to be found. This is the critical step in the expansion move algorithm of
This section is organized as follows. The construction of Gα for a given f (or P) and a is described. Then cuts C on Gα are shown to correspond in a natural way to labelings fC which are within one α-expansion move of f. Then, based on a number of simple properties, a class of elementary cuts is defined. Theorem 5.4 shows that elementary cuts are in one to one correspondence with labelings that are within one α-expansion of f, and also that the cost of an elementary cut is |C|=E(fC). A corollary from this theorem states he main result that the desired labeling {circumflex over (f)} is fC where C is a minimum cut on Gα.
The structure of the graph is illustrated in FIG. 5. For legibility, this figure shows the case of a 1-D image. The set of vertices includes the two terminals a 202 and {overscore (α)} 204, as well as all image pixels p∈P 206. In addition, for each pair of neighboring pixels {p,q}∈N separated in the current partition (i.e. such that fp≠fq, an auxiliary vertex a{p,q} is created. Auxiliary nodes are introduced at the boundaries between partition sets Pl for l∈L. Thus, the set of vertices is
Each pixel p∈P is connected to the terminals α and {overscore (α)} by t-links tpα208 and tp{overscore (α)}210, respectively. Each pair of neighboring pixels {p,q}∈N which are not separated by the partition P (i.e. such that fp=fq) is connected by an n-link; e{p,q}212. For each pair of neighboring pixels {p,q}∈N such that fp≠fq a triplet of edges ε{p,q}={e{p,a},e{a,q},ta{overscore (α)}} is created where a=a{p,q} is the corresponding auxiliary node. The edges e{p,a} and e{a,q} connect pixels p and q to a{p,q} and the t-link ta{overscore (α)} connects the auxiliary node a{p,q} to the terminal {overscore (α)}. The set of all edges may be written as
The weights are assigned to the edges according to, Table 2.
TABLE 2 | ||
edge | weight | for |
tp{overscore (α)} | ∞ | p ∈ Pα |
tp{overscore (α)} | Dp(fp) | p ∉ Pα |
tpα | Dp(α) | p ∈ P |
e{p,a} | V(fp,α) | {p,q} ∈ N, fp ≠ fq |
e{a,q} | V(α,fq) | {p,q} ∈ N, fp ≠ fq |
ta{overscore (α)} | V(fp,fq) | {p,q} ∈ N, fp ≠ fq |
e{p,q} | V(fp,α) | {p,q} ∈ N, fp = fq |
Any cut C on Gαmust sever (include) exactly one t-link for any pixel p∈P. This defines a natural labeling fC corresponding to a cut C on Gα. Formally,
In other words, a pixel p is assigned label α if the cut C separates p from the terminal α, while p is assigned its old label fp if C separates p from {overscore (α)}. Note that for p∉Pα the terminal {overscore (α)} represents labels assigned to pixels in the initial labeling f. Clearly this leads to:
Lemma 5.1 A labeling fC corresponding to a cut C on Gα is one α-expansion away from the initial labeling f.
It is also easy to show that a cut C severs an n-link e{p,q} between neighboring pixels {p,q}∈N such that fp=fq if and only if C leaves the pixels p and q connected to different terminals. In other words, property 4.2 holds when "{overscore (α)}" is substituted for "β". This as property is referred to as 4.2({overscore (α)}). Analogously, that property 4.2 and equation 16 establish lemma 4.3 for the n-links e{p,q} in Gα.
Consider now the set of edges ε{p,q} corresponding to a pair of neighboring pixels {p,q}∈N such that fp≠fq. In this case, there are several different ways to cut these edges even when the pair of severed t-links at p and q is fixed. However, a minimum cut C on Gα is guaranteed to sever the edges in ε{p,q} depending on what t-links are cut at the pixels p and q.
Property 5.2 If {p,q}∈N and fp≠fq then a minimum cut C on Gα satisfies:
Property 5.2(a) results from the fact that no subset of C is a cut. The others follow from the minimality of |C| and the fact that |e{p,a}|, |e{a,q}| and |ta{overscore (α)} satisfy the triangle inequality so that cutting any one of them is cheaper than cutting the other two together.
Lemma 5.3 If If {p,q}∈N and fp≠fq then the minimum cut C on Gα satisfies:
Proof: The equation follows from property 5.2, equation (16), and the edge weights. For example,
If tp{overscore (α)},tq{overscore (α)}∈C then C∩ε{p,q}=|ta{overscore (α)}=V(fp,fq) At the same time, (16) implies that fpC=fp and fpC=fq. Note that the right penalty V is imposed whenever fpC≠fqC, due to the auxiliary pixel construction.
Property 4.2 ({overscore (α)}) holds for any cut, and property 5.2 holds for a minimum cut. However, there can be other cuts; besides the minimum cut that satisfy both properties. An elementary cut on Gα is defined as a cut that satisfies properties 4.2 ({overscore (α)}) and 5.2.
Theorem 5.4 Let Gα be constructed as above given f and α. Then there is a one to one correspondence between elementary cuts on Gα and labelings within one α-expansion of f. Moreover, for any elementary cut C, |C|=E(fC) is the result. Proof: An elementary cut C is uniquely determined by the corresponding labeling fC. The label fpC at the pixel p determines which of the t-links to p is in C. Property 4.2 ({overscore (α)}) shows which n-links e{p,q} between pairs of neighboring pixels {p,q} such that fp=fq should be severed. Similarly, property 5.2 determines which of the links in ε{p,q} corresponding o {p,q}∈N such that fp≠fq should be cut.
The cost of an elementary cut C is
It is easy to show that for any pixel p∈P, the result is
Lemmas 4.3 and 5.3 hold for elementary cuts, since they were based on properties 4.2 and 5.2. Thus, the total cost of a elementary cut C is
Therefore, |C|=E(fC).
Our main result is a simple consequence of this theorem, since the minimum cut is an elementary cut.
Corollary 5.5 The lowest energy labeling within a single α-expansion move from f is {circumflex over (f)}=fC, where C is the minimum cut on Gα.
Next, a label is chosen, block 252. The chosen label is expanded, block 254. That is, any pixel is allowed to take on the chosen label. Then the best solution is determined using the minimum graph cut method, block 256. Each link in the graph is assigned a weight according to Table 2 and cuts are made to determine the best solution as described above. In a preferred embodiment of the invention, the method then determines if any more improvements may be made, block 258. If yes, the method repeats until no further improvements may be made and then the process ends, block 260. In an alternative embodiment of the invention, the iteration continues only until an acceptable solution is obtained.
Optimality Properties
Any local minimum generated by the expansion moves algorithm of the present invention is within a known factor from the global optimum. This algorithm works in case of metric V. The swap move algorithm can be applied to a wider class of semi-metric V's but, unfortunately, it does not have any guaranteed optimality properties. A provably good solution can be obtained even for semi-metric V by approximating such V's with a simple Potts metric.
The Expansion Move Algorithm
A local minimum when expansion moves are allowed is within a known factor of the global minimum. This factor, which can be as small as 2, will depend on V. Specifically, let
be the ratio of the largest non zero value of V to the smallest non zero value of V. Note that c is well defined since V(α,β)≠0 for α≠β according to the metric properties (2) and (3). If Vp,q's are different for neighboring pairs p,q
Theorem 6.1 Let {circumflex over (f)} be a local minimum when the expansion moves are allowed and f* be the globally optimal solution. Then E({circumflex over (f)})≦2cE(f*)
Proof: Fix some α∈L and let
A labeling fα can be produced within one α-expansion move from {circumflex over (f)} as follows:
The key observation is that since {circumflex over (f)} is a local minimum if expansion moves are allowed,
E({circumflex over (f)})≦E(fα). (23)
Let S be a set consisting of any number of pixels in P and any number of pairs of neighboring pixels in N. E(f|S) is defined to be a restriction of the energy of labeling f to the set S:
Let Iα be the set of pixels and pairs of neighboring pixels contained inside Pα. Also, let Bα be the set of pairs of neighboring pixels on the boundary of Pα and Oα be the set of pixels and pairs of neighboring pixels contained outside of Pα. Formally,
The following three facts hold:
Equations 28 and 29 are obvious from the definitions in (22) and (21). Equation (30) holds because for any {p,q}∈Bα there is V(fpα,fqα)≦cV(f*p,f*q)≠0.
Since Iα∪Bα∪Oα includes all pixels in P and all neighboring pairs of pixels in N, both sides of (23) can be expanded to get:
Using (28), (29) and (30) the following equation is derived from the equation above:
To get the bound on the total energy, equation (32) is summed over all labels α∈L:
Let
Observe that for every {p,q}∈B, the term V({circumflex over (f)}p,{circumflex over (f)}q)=E({circumflex over (f)}|{p,q}) appears twice on the left side of (33), once in E({circumflex over (f)}|Bα) for α=f*p and once in E({circumflex over (f)}|Bα) for α=f*q. Similarly every V(f*p,f*q*)=E(f*|{p,q}) appears 2c times on the right side of (33). Therefore equation (33) can be rewritten to get the bound of 2c:
It can be shown that any algorithm that is within a factor of 2 for the Potts model is within a factor of 2c for an arbitrary metric V.
Approximating a Semi-metric
The expansion algorithm of the present invention can be used to get an answer within a factor of 2c from the optimum of energy (4) even when V is a semi-metric. Actually just V(α,β)≧0 and V(α,β)=0⇄α=β are needed. Here c is the same as in Theorem 6.1. This c is still well defined for a semi-metric. Suppose that penalty V inside the definition of energy E in (4) is a semi-metric. Let r be any real number in the interval [m,M] where
Define a new energy based on the Potts interaction model
Theorem 6.2 If {circumflex over (f)} is a local minimum of EP given the expansion moves and f* is the global minimum of E(f) then E({circumflex over (f)})≦2cE(f*).
Proof: Suppose f°C is the global minimum of EP. Then
where the second inequality follows from Theorem 6.1. Note that c=M/m.
Thus to find an answer within a fixed factor from the global minimum for a semi-metric V, one can take a local minimum {circumflex over (f)} given the expansion moves for EP as defined above. Note that that such an {circumflex over (f)} is not a local minimum of E(f) given the expansion moves. In practice, it has been found find that local minimum given the swap moves gives empirically better results than using {circumflex over (f)}. In fact, the estimate {circumflex over (f)} can be used as a good starting point for the swap algorithm. In this case, the swap move algorithm will also generate a local minimum whose energy is within a known factor from the global minimum.
The Potts Model
An interesting special case of the energy in equation (4) arises when V is given by the Potts model
In this case, discontinuities between any pair of labels are penalized equally. This is in some sense the simplest discontinuity preserving model and it is especially useful when the labels are unordered or the number of labels is small. The Potts interaction penalty Vp,q=u{p,q}·T(fp≠fq) is a metric; in this case c=1 and the expansion algorithm of the present invention gives a solution that is within a factor of 2 of the global minimum. Note that by definition c≧1, so this is the energy function with the best bound.
The Potts model energy minimization problem is closely related to a known combinatorial optimization problem called the multiway cut problem. The Potts model energy minimization problem can be reduced to the multiway cut problem. More precisely, it can be proved that the global minimum of the Potts model energy EP can be computed by finding the minimum cost multiway cut on an appropriately constructed graph. In order to efficiently compute the global minimum of EP, a certain class of multiway cut problems that are known to be NP-hard could also be solved. Minimizing EP is NP-hard, and so is minimizing the energy in (4).
The multiway cut problem is defined on a graph G=(V,E), with non-negative edge weights, with a set of terminal vertices L⊂V. A subset of the edges C⊂ε is called a multiway cut if the terminals are completely separated in the {induced} graph G(C)=(V, ε-C). It is also required that no proper subset of C separates the terminals in G(C). The cost of the multiway cut C is denoted by |C| and equals the sum of its edge weights. The multiway cut problem is to find the minimum cost multiway cut. The multiway cut problem is NP-: complete. The multiway cut problem is a generalization of the standard two-terminal graph cut problem.
The Potts Model and the Multiway Cut Problem
The problem of minimizing the Potts energy EP(f) can be solved by computing a minimum cost multiway cut on a certain graph. Given V=P∪L, means that G contains two types of vertices: p-vertices (pixels) and l-vertices (labels). Note that l-vertices will serve as terminals or the multiway cut problem. Two p-vertices are connected by an edge if and only if the corresponding pixels are neighbors in the neighborhood system N. The set εN consists of the edges between p-vertices, called n-links. Each n-link {p,q}∈εN is assigned a weight w{p,q}=u{p,q}.
Each p-vertex is connected by an edge to each l-vertex. An edge {p,l} that connects a p-vertex with a terminal (an l-vertex) will be called a t-link and the set of all such edges will be denoted by ετ. Each t-link {p,l} in ετ is assigned a weight w{p,l}=Kp-Dp(l); where Kp>maxlDp(l) is a constant that is large enough to make the weights positive. The edges of the graph are ε=εN∪ετ.
It is easy to see that there is a one-to-one correspondence between multiway cuts and labelings. A multiway cut C corresponds to the labeling fC which assigns the label l to all pixels p which are t-linked to the l-vertex in G(C). An example of a multiway cut and the corresponding image partition (labeling) is given in FIG. 9b. The multiway cut corresponds to a unique partition (labeling) of image pixels.
Theorem 7.1 If C is a multiway cut on Gα then |C|=EP(fC) plus a constant.
Corollary 7.2 If C is a minimum cost multiway cut on G, then fC minimizes EP.
While the multiway cut problem is known to be NP-complete if there are more than 2 terminals, there is a fast approximation algorithm. This algorithm works as follows. First, for each terminal l∈L, it finds an isolating two-way minimum cut C(l) that separates l from all other terminals. This is just the standard graph cut problem. Then the algorithm generates a multiway cut C=Ul≠l
However, the isolation heuristic algorithm suffers from two problems that limits its applicability to the energy minimization problem.
First, the algorithm will assign many pixels a label that is chosen essentially arbitrarily. Note that the union of all isolating cuts Ul∈LC(l) may leave some vertices disconnected from any terminal. The multiway cut C=Ul≠l
Second, while the multiway cut C produced is close to optimal, this does not imply that the resulting labeling fC is close to optimal. Formally, write theorem 7.1 as |C|=EP(C)+K (the constant K results from the Kp's. The isolation heuristic gives a solution Ĉ such that |Ĉ|≦2|C*|, where C* is the minimum cost multiway cut. Thus, EP(Ĉ)+K≦2(EP(C*)+K), so EP(Ĉ)≦2EP(C*)+K. As a result, the isolation heuristic algorithm does not produce a labeling whose energy is within a constant factor of optimal.
Minimizing the Potts Energy is NP-hard
For an arbitrary fixed graph G=(V,E), it is possible to construct an instance of minimizing EP(f) where the optimal labeling f* determines a minimum multiway cut on G. This will prove that a polynomial-time method for finding f* would provide a polynomial-time algorithm for finding the minimum cost multiway cut, which is known to be NP-hard.
The energy minimization problem takes as input a set of pixels P, a neighborhood relation N and a label set L, as well as a set of weights u{p,q} and a function Dp(l). The problem is to find the labeling f* that minimizes the energy EP(f) given in equation (36).
Let G=(V,ε) be an arbitrary weighted graph with terminal vertices {t1, . . . , tk}⊂V and edge weights w{p,q}. The energy minimization using P=V, N=E, and u{p,q}=w{p,q}. The label set will be L={1, . . . , k}. Let K be a constant such that K>EP(f*); for example, K can be selected to be the sum of all w{p,q}. The function Dp(l) will force f*(tj)=j; if p=tj is a terminal vertex,
For a non-terminal vertex p all labels are equally good,
A labeling f is defined to be feasible if the set of pixels labeled j by f forms a connected component that includes tj. Feasible labelings obviously correspond one-to-one with multiway cuts.
Theorem 7.3 The labeling f* is feasible, and the cost of a feasible labeling is the cost of the corresponding multiway cut.
Proof: To prove that f* is feasible, suppose that there were a set S of pixels that f* labeled j which were not part of the component containing tj. A labeling could then be obtained having lower energy by switching this set to the label of some pixel on the boundary of S. The energy of a feasible labeling f is
which is the cost of the multiway cut corresponding to f.
This shows that minimizing the Potts model energy EP(f) on an arbitrary P and N is intractable. In computer vision, however, P is usually a planar grid, and combinatorial problems that are intractable on arbitrary graphs sometimes become tractable on the plane or grid.
The energy minimization problem is intractable even when restricted to a planar grid. The reduction is from a special case of the multiway cut problem, where G is a planar graph with degree 11 and all the edges have weight 1, which is NP-hard. G must be embedded in a grid of pixels, which happens in two stages. In the first stage, G is converted into a planar graph of degree 4. In the second stage, this graph is embedded in the grid. This embedding can be done in polynomial time; after it is done, each vertex v E G corresponds to a connected set of pixels S(v) in the grid, and the adjacency relationships among vertices in G has been preserved.
The proof now proceeds along the same lines as theorem 7.3, except for three subtleties. First, for every vertex v all pixels in S(v) are given the same label. This is ensured by making the edge weights K between adjacent pixels in S(v). Second, there are gaps when G is embedded in the grid. This can be solved by adding additional "grid pixels", which D forces to have the extra label 0 (D will prevent non-grid pixels from having label 0 by making Dp(0)=K). The edge weights are taken between grid pixels and non-grid pixels to be one. The cost of a feasible labeling is now the cost of the corresponding multiway cut plus a constant. Third, the constant K>EP(f*) must be now chosen more carefully.
Experimental Results
The algorithms of the present invention were applied to image restoration and visual correspondence and achieved promising results.
The experimental results are from visual correspondence for stereo and motion. In visual correspondence, two images are taken at the same time from different view points for stereo, and are also taken from the same view point but at different times for motion. For each pixel in the first image there is a corresponding pixel in the second image which is a projection along the line of sight of the same real world scene element. The difference in the coordinates of the corresponding points is called the disparity. In stereo the disparity is usually one-dimensional because corresponding points lie along epipolar lines. In motion the disparity is usually a two-dimensional quantity. The disparity varies smoothly everywhere except at object boundaries, and corresponding points are expected to have similar intensities. The correspondence problem can be formulated as the energy minimization problem
Here P is the set of all pixels in the first image, Ip is the intensity of pixel p in the first image, Iq' is the intensity of pixel q in the second image, and p+fp stands for the pixel with coordinates of p shifted by disparity fp. The data penalty D is small if there is a small difference between Ip and i'p+f
For the experiments, three energy functions were used. The first energy function, called EQ, uses the truncated quadratic V(fp,fq)=min(K,|fp,fq|2) (for some constant K) as its smoothness term. This choice of V does not obey the triangle inequality, so EQ was minimized using the swap algorithm of the present invention.
The second (EP) and the third (EL) energy functions use, correspondingly, the Potts model and the truncated L2 distance as their smoothness penalty V. Both of these obey the triangle inequality and EP and EL were minimized with the expansion algorithm of the present invention.
Data Term
If the pixels p and q correspond, they are assumed to have similar intensities Ip and I'q. Thus (Ip-I'q)2 is frequently used as a penalty for deciding that p and q correspond. This penalty has a heavy weight unless Ip≈I'q. However there are special circumstances when corresponding pixels have very different intensities due to the effects of image sampling. Suppose that the true disparity is not an integer. If a pixel overlaps a scene patch with high intensity gradient, then the corresponding pixels may have significantly different intensities.
For stereo, a technique is used to develop a Dp that is insensitive to image sampling. First it was measured how well p fits into the real valued range of disparities
by
Fractional values I'p+x are the result by linear interpolation between discrete pixel values. For symmetry,
is also measured. Cfwd(p,d) and Crev(p,d) can be computed with just a few comparisons. The final measure is
Here Const is used to make the measure more robust.
Static Cues
In this section, how to choose different Vp,q for each pair of interacting pixels {p,q} to take advantage of contextual information is described. For simplicity, consider the case of the Potts model, i.e. Vp,q=u{p,q}·T(fp≠fq). The intensities of pixels in the first image contain information that can significantly influence the assessment of disparities without even considering the second image. For example, two neighboring pixels p and q are much more likely to have the same disparity if it is known that I(p)≈I(q). Most methods for computing correspondence do not make use of this kind of contextual information.
Contextual information can be incorporated into the framework by allowing u{p,q} to vary depending on their intensities Ip and Iq. Let
Each u{p,q} represents a penalty for assigning different disparities to neighboring pixels p and q. The value of the penalty u{p,q} should be smaller for pairs {p,q} with larger intensity differences |Ip-Iq|. In practice, an empirically selected decreasing function U(•) is used. Note that instead of (19) the coefficients u{p,q} could also be set according to an output of an edge detector on the first image. For example, u{p,q} can be made small for pairs {p,q} where an intensity edge was detected and large otherwise. Segmentation results can also be used.
The following example shows the importance of contextual information. Consider the pair of synthetic images below, with a uniformly white rectangle in front of a black background.;
There is a one pixel horizontal shift in the location of the rectangle, and there is no noise. Without noise, the problem of estimating f is reduced to minimizing the smoothness term Esmooth(f) under the constraint that pixel p can be assigned disparity d only if Ip=I'p+d.
If u{p,q} is the same for all pairs of neighbors {p,q} then Esmooth(f) is minimized at one of the labeling shown in the picture below. Exactly which labeling minimizes Esmooth(f) depends on the relationship between the height of the square and the height of the background.
Suppose now that the penalty u{p,q} is much smaller if Ip≠Iq than it is if Ip=Iq. In this case the minimum of Esmooth(f) is achieved at the disparity configuration shown in the picture below. This result is much closer to human perception.
Real Stereo Imagery With Ground Truth
For this stereo pair, EP was used. The results were compared against annealing and normalized correlation. For normalized correlation, parameters which give the best statistics were chosen. Several different annealing variants were implemented. The one that gave the best performance was used. This was the Metropolis sampler with a linearly decreasing temperature schedule. To give it a good starting point, simulated annealing was initialized with the results from normalized correlation. In contrast, for the algorithms of the present invention, the starting point is unimportant. The results differ by less than 1% of image pixels from any starting point that was tried.
FIGS. 10(c), and (d) show the results of the swap and expansion algorithms for λ=20. FIGS. 10(e) and (f) show the results of normalized correlation and simulated annealing.
The table below summarizes the errors made by the algorithms. In approximately 20 minutes, simulated annealing reduces the total errors normalized correlation makes by about one fifth and it cuts the number of ±1 errors in half. It makes very little additional progress in the rest of 19 hours that it was run. The expansion, swap, and jump algorithms of the present invention make approximately 3 times fewer ±1 errors and approximately 5 times fewer total errors compared to normalized correlation.
The expansion and swap algorithms of the present invention perform similarly to each other. The observed slight difference in errors is quite insignificant (less than one percent). At each cycle the order of labels to iterate over is chosen randomly. Another run of the algorithms might give slightly different results, where expansion algorithm might do better than the swap algorithm. In general, very slight variation between different runs of an algorithm was observed. However the difference in the running time is significant. On average the expansion algorithm converges 3 times as rapidly as the swap algorithm.
% of errors | |||
algorithm | % total errors | >±1 | running time |
expansion | 7.6 | 2.1 | 106 sec |
algorithm | |||
swap algorithm | 7.0 | 2.0 | 300 sec |
simulated | 20.3 | 5.0 | 1200 sec |
annealing | |||
normalized | 24.7 | 10.0 | 5 sec |
correlation | |||
Thus the Edata term of the energy function has very similar values for the methods of the present invention and simulated annealing he methods of the present invention quickly reduce the smoothness energy to around 16,000, while the best simulated annealing can produce in 19 hours is around 30,000, nearly twice as bad.
The expansion algorithm gives a convergence curve significantly steeper than the other curves. In fact the expansion algorithm makes 99% of the progress in the first iteration.
The algorithms appear to be quite stable in the choice of parameter λ. For example the table in
SRI Tree Stereo Pair
In the well-known SRI stereo pair whose left image is shown in FIG. 13(a) the number of disparities is larger, and EP does not work as well. FIG. 13(b) and (c) compares the results of minimizing EP and EL. Notice that there are fewer disparities found in FIG. 13(b), since the piecewise constant prior tends to produce large regions with the same disparity.
Motion
FIG. 14(a) shows one image of a motion sequence where a cat moves against moving background. This is a difficult sequence because the cat's motion is non-rigid. FIGS. 14(b) and (c) show the horizontal and vertical motions detected with the swap algorithm of the present invention. Notice that the cat has been accurately localized. Even the tail and parts of the legs are clearly separated from the background motion.
Bayesian Labeling of Markov Random Fields
In this appendix it is shown that minimization of the energy function in (1) is equivalent to computing the maximum a posteriori estimate of a Markov Random Field (MRF).
Let P be a set of sites, La set of labels, and N={{p,q}|p,q∈P} a neighborhood system on P. Let F=F1, . . . , Fn be a set of random variables defined on P. Each Fp takes values in the label set L. A particular realization of the field will be denoted by f={fp|p∈P}, which is also called a configuration of the field F. As usual, P(Fp=fp) will be abbreviated by P(fp). F is said to be a Markov Random Field (MRF) if:
(i) P(f)>0 ∀f∈F.
(ii) P(fp |fP-{p})=P(fp|fN
where P-{p} denotes set difference, fN
The easiest way to specify an MRF is by joint distribution. This theorem proves the equivalence between MRFs and Gibbs random fields.
Before defining Gibbs random fields, a clique needs to be defined. A set of sites is called a clique if each member of the set is a neighbor of all the other members. A Gibbs random field can be specified by the Gibbs distribution:
where C is the set of all cliques, Z is the normalizing constant, and {Vc(f)} are functions from a labeling to non-negative reals, called the clique potential functions.
Thus to specify an MRF, the clique potential functions need to be specified. A first order MRF is considered, which means that for all cliques of size larger than two the potential functions are zero, and for the cliques of size two the potential functions are specified by
This defines an MRF with the joint distribution:
In general, the field F is not directly observable in the experiment. A popular way to estimate its realized configuration {circumflex over (f)} based on an observation d is the maximum a posteriori (MAP) estimation. Using Bayes rule, the posterior probability can be written as
Thus the MAP estimate f* is equal to
Assume that he observation dp at each pixel is independent and that
where Cp is the normalizing constant, and Dp was defined in Section 1. Then the likelihood can be written as
Writing out p(d) and p(d|f) with the above assumptions results in
which is the general form of the energy function being minimized.
It is to be understood that the above-described embodiments are simply illustrative of the principles of the invention. Various and other modifications and changes may be made by those skilled in the art which will embody the principles of the invention and fall within the spirit and scope thereof.
Zabih, Ramin D., Veksler, Olga, Boykov, Yuri
Patent | Priority | Assignee | Title |
10268930, | Jul 22 2015 | Canon Kabushiki Kaisha | Image processing system and method for identifying content within image data using markov random fields |
10282874, | Sep 17 2014 | CIRCONUS, INC.; CIRCONUS, INC | Efficient time-series histograms |
10354391, | Oct 27 2016 | Method and system for multi-label image segmentation | |
10467245, | Apr 27 2010 | Cornell University | System and methods for mapping and searching objects in multidimensional space |
6973212, | Sep 01 2000 | Siemens Medical Solutions USA, Inc | Graph cuts for binary segmentation of n-dimensional images from object and background seeds |
7212201, | Sep 23 1999 | New York University | Method and apparatus for segmenting an image in order to locate a part thereof |
7430339, | Aug 09 2004 | Microsoft Technology Licensing, LLC | Border matting by dynamic programming |
7444019, | Feb 18 2004 | Siemens Medical Solutions USA, Inc | System and method for GPU acceleration of push-relabel algorithm on grinds |
7460709, | Jan 23 2004 | Siemens Medical Solutions USA, Inc | System and method for multi-label image segmentation |
7486820, | Jan 06 2005 | Siemens Healthcare GmbH | System and method for multilabel random walker segmentation using prior models |
7519220, | Nov 15 2004 | Siemens Medical Solutions USA, Inc | GPU accelerated isoperimetric algorithm for image segmentation, digital photo and video editing |
7542604, | Aug 26 2004 | Siemens Medical Solutions USA, Inc | System and method for image segmentation by solving an inhomogenous dirichlet problem |
7565010, | Jan 06 2005 | Siemens Medical Solutions USA, Inc | System and method for image segmentation by a weighted multigrid solver |
7570804, | Dec 07 2004 | Electronics and Telecommunications Research Institute | Apparatus and method for determining stereo disparity based on two-path dynamic programming and GGCP |
7630549, | Nov 15 2004 | Siemens Medical Solutions USA, Inc | GPU accelerated multi-label digital photo and video editing |
7672516, | Mar 21 2005 | Siemens Medical Solutions USA, Inc | Statistical priors for combinatorial optimization: efficient solutions via graph cuts |
7715627, | Mar 25 2005 | Siemens Medical Solutions USA, Inc | Automatic determination of the standard cardiac views from volumetric data acquisitions |
7835577, | Jan 18 2005 | Siemens Medical Solutions USA, Inc | System and method for multi-label image segmentation of digital photographs |
7889924, | Apr 10 2006 | Siemens Medical Solutions USA, Inc | Globally optimal uninitialized graph-based rectilinear shape segmentation |
7893939, | Sep 23 1999 | New York University | Method and apparatus for segmenting an image in order to locate a part thereof |
7925089, | Sep 18 2007 | Microsoft Technology Licensing, LLC | Optimization of multi-label problems in computer vision |
8009886, | Jan 03 2008 | General Electric Company | System and method for image registration |
8041114, | Jun 15 2007 | Microsoft Technology Licensing, LLC | Optimizing pixel labels for computer vision applications |
8050498, | Jul 21 2006 | Adobe Inc | Live coherent image selection to differentiate foreground and background pixels |
8103068, | Sep 29 2006 | Cornell Research Foundation, Inc | Methods and systems for reconstruction of objects |
8121407, | Mar 17 2008 | Adobe Inc | Method and apparatus for localized labeling in digital images |
8144954, | Nov 08 2007 | D4D Technologies, LLC | Lighting compensated dynamic texture mapping of 3-D models |
8200014, | Jul 21 2006 | Adobe Inc | Live coherent image selection |
8213726, | Jun 19 2009 | Microsoft Technology Licensing, LLC | Image labeling using multi-scale processing |
8441481, | Sep 23 1999 | New York University | Method and apparatus for segmenting an image in order to locate a part thereof |
8452087, | Sep 30 2009 | Microsoft Technology Licensing, LLC | Image selection techniques |
8503783, | Sep 18 2007 | Microsoft Technology Licensing, LLC | Optimization of multi-label problems in computer vision |
8532355, | Nov 08 2007 | D4D Technologies, LLC | Lighting compensated dynamic texture mapping of 3-D models |
8533139, | Sep 06 2011 | Czech Technical University in Prague, Faculty of Electrical Engineering | Optimizing computation of minimum cut in graphs with grid topology |
8542923, | Jul 21 2006 | Adobe Inc | Live coherent image selection |
8655069, | Mar 05 2010 | Microsoft Technology Licensing, LLC | Updating image segmentation following user input |
8854363, | Sep 23 1999 | New York University | Method and apparatus for segmenting an image in order to locate a part thereof |
8860838, | Sep 24 2008 | Nikon Corporation | Automatic illuminant estimation and white balance adjustment based on color gamut unions |
9013596, | Sep 24 2008 | Nikon Corporation | Automatic illuminant estimation that incorporates apparatus setting and intrinsic color casting information |
9025043, | Sep 24 2008 | Nikon Corporation | Image segmentation from focus varied images using graph cuts |
9076216, | Aug 15 2011 | ALMALENCE, INC | Method of image segmentation |
9082176, | Oct 25 2010 | Samsung Electronics Co., Ltd.; SAMSUNG ELECTRONICS CO LTD | Method and apparatus for temporally-consistent disparity estimation using detection of texture and motion |
9118880, | Sep 24 2008 | Nikon Corporation | Image apparatus for principal components analysis based illuminant estimation |
9317536, | Apr 27 2010 | Cornell University | System and methods for mapping and searching objects in multidimensional space |
9582891, | Sep 23 1999 | New York University | Method and apparatus for segmenting an image in order to locate a part thereof |
9916667, | Jul 18 2014 | Samsung Electronics Co., Ltd. | Stereo matching apparatus and method through learning of unary confidence and pairwise confidence |
Patent | Priority | Assignee | Title |
6507661, | Apr 20 1999 | NEC Corporation | Method for estimating optical flow |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Aug 30 2000 | Cornell Research Foundation, Inc. | (assignment on the face of the patent) | / | |||
Dec 05 2000 | VEKSLER, OLGA | Cornell Research Foundation, Inc | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 011414 | /0241 | |
Dec 05 2000 | BOYKOV, YURI | Cornell Research Foundation, Inc | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 011414 | /0241 | |
Dec 13 2000 | ZABIH, RAMIN D | Cornell Research Foundation, Inc | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 011414 | /0241 | |
May 11 2009 | Cornell University | NATIONAL SCIENCE FOUNDATION | CONFIRMATORY LICENSE SEE DOCUMENT FOR DETAILS | 023030 | /0986 |
Date | Maintenance Fee Events |
Sep 04 2007 | ASPN: Payor Number Assigned. |
Dec 03 2007 | M2551: Payment of Maintenance Fee, 4th Yr, Small Entity. |
Dec 10 2007 | REM: Maintenance Fee Reminder Mailed. |
Dec 01 2011 | M2552: Payment of Maintenance Fee, 8th Yr, Small Entity. |
Nov 19 2015 | STOL: Pat Hldr no Longer Claims Small Ent Stat |
Dec 01 2015 | M1553: Payment of Maintenance Fee, 12th Year, Large Entity. |
Date | Maintenance Schedule |
Jun 01 2007 | 4 years fee payment window open |
Dec 01 2007 | 6 months grace period start (w surcharge) |
Jun 01 2008 | patent expiry (for year 4) |
Jun 01 2010 | 2 years to revive unintentionally abandoned end. (for year 4) |
Jun 01 2011 | 8 years fee payment window open |
Dec 01 2011 | 6 months grace period start (w surcharge) |
Jun 01 2012 | patent expiry (for year 8) |
Jun 01 2014 | 2 years to revive unintentionally abandoned end. (for year 8) |
Jun 01 2015 | 12 years fee payment window open |
Dec 01 2015 | 6 months grace period start (w surcharge) |
Jun 01 2016 | patent expiry (for year 12) |
Jun 01 2018 | 2 years to revive unintentionally abandoned end. (for year 12) |